Skip to content

V
vscode

xxadev / vscode
与 Fork 源项目一致

从无法访问的项目Fork

通知 2
Star 0
Fork 0
V
vscode

体验新版 GitCode，发现更多精彩内容 >>

提交 76b6b177 编写于 作者: P Phil Price
浏览文件
操作

Add line by line parser for merge conflicts. 

Prior implementation in mergeConflictParserLegcacy which does a one shot regular expression.
上级 59fbe319
隐藏空白更改
内联 并排
Showing with 204 addition and 94 deletion
extensions/merge-conflict/src/mergeConflictParser.ts
浏览文件 @ 76b6b177
......@@ -5,125 +5,106 @@
import * as vscode from 'vscode';
import * as interfaces from './interfaces';
import { DocumentMergeConflict } from './documentMergeConflict';
import * as vm from 'vm';
const startMarker = '<<<<<<<';
const splitterMarker = '=======';
const endMarker = '>>>>>>>';
interface IPartialMergeConflictDescriptor {
currentHeader: vscode.Range;
splitter?: vscode.Range;
incomingFooter?: vscode.Range;
}
export class MergeConflictParser {
static scanDocument(document: vscode.TextDocument): interfaces.IDocumentMergeConflict[] {
// Conflict matching regex, comments are in the format of "description - [group index] group name"
// Premise is: Match the current change (<<<<<<), match anything up to the splitter (======) then
// match anything up to the incoming change (>>>>>>), this leaves some oddities with newlines not being
// pulled into the "body" of each change, DocumentMergeConflict.applyEdit will deal with these cases
// and append newlines when needed
// Scan each line in the document, we already know there is atleast a <<<<<<< and
// >>>>>> marker within the document, we need to group these into conflict ranges.
const conflictMatcher = new RegExp([
/(^<<<<<<<\s(.+)\r?\n)/, // "Current" conflict header - [1] entire line, [2] name
/([\s\S]*?)/, // "Current" conflict body - [3] body text
/(^=======\r?\n)/, // Splitter - [4] entire line
/([\s\S]*?)/, // Incoming conflict body - [5]
/(^>>>>>>>\s(.+)\r?\n)/ // Incoming conflict header - [6] entire line, [7] name
].map(r => r.source).join(''), 'mg');
let currentConflict: IPartialMergeConflictDescriptor | null = null;
const offsetGroups = [1, 3, 4, 5, 6]; // Skip inner matches when calculating length
const conflictDescriptors: interfaces.IDocumentMergeConflictDescriptor[] = [];
let text = document.getText();
let sandboxScope = {
result: [],
conflictMatcher,
text: text
};
const context = vm.createContext(sandboxScope);
const script = new vm.Script(`
let match;
while (match = conflictMatcher.exec(text)) {
// Ensure we don't get stuck in an infinite loop
if (match.index === conflictMatcher.lastIndex) {
conflictMatcher.lastIndex++;
for (let i = 0; i < document.lineCount; i++) {
const line = document.lineAt(i);
if (line.text.startsWith(startMarker)) {
if (currentConflict !== null) {
// Error, we should not see a startMarker before we've seen an endMarker
currentConflict = null;
break;
}
result.push(match);
}`);
currentConflict = {
currentHeader: line.range
};
}
else if (line.text.startsWith(splitterMarker)) {
try {
// If the regex takes longer than 1s consider it dead
script.runInContext(context, { timeout: 1000 });
}
catch (ex) {
return [];
}
if (currentConflict === null) {
continue; // Ignore
}
return sandboxScope.result.map(match => new DocumentMergeConflict(document, MergeConflictParser.matchesToDescriptor(document, match, offsetGroups)));
}
currentConflict.splitter = line.range;
}
else if (line.text.startsWith(endMarker)) {
if (currentConflict === null) {
continue; // Ignore
}
static containsConflict(document: vscode.TextDocument): boolean {
if (!document) {
return false;
}
currentConflict.incomingFooter = line.range;
// TODO: Ask source control if the file contains a conflict
let text = document.getText();
return text.includes('<<<<<<<') && text.includes('>>>>>>>');
}
let completeDescriptor = MergeConflictParser.completePartialMergeDescriptor(currentConflict);
static matchesToDescriptor(document: vscode.TextDocument, match: RegExpExecArray, offsets?: number[]): interfaces.IDocumentMergeConflictDescriptor {
if (completeDescriptor !== null) {
conflictDescriptors.push(completeDescriptor);
}
var item: interfaces.IDocumentMergeConflictDescriptor = {
range: new vscode.Range(document.positionAt(match.index), document.positionAt(match.index + match[0].length)),
current: {
name: match[2],
header: this.getMatchPositions(document, match, 1, offsets),
content: this.getMatchPositions(document, match, 3, offsets),
},
splitter: this.getMatchPositions(document, match, 4, offsets),
incoming: {
name: match[9],
header: this.getMatchPositions(document, match, 6, offsets),
content: this.getMatchPositions(document, match, 5, offsets),
currentConflict = null;
}
};
return item;
}
static getMatchPositions(document: vscode.TextDocument, match: RegExpExecArray, groupIndex: number, offsetGroups?: number[]): vscode.Range {
// Javascript doesnt give of offsets within the match, we need to calculate these
// based of the prior groups, skipping nested matches (yuck).
if (!offsetGroups) {
offsetGroups = match.map((i, idx) => idx);
}
let start = match.index;
for (var i = 0; i < offsetGroups.length; i++) {
let value = offsetGroups[i];
return conflictDescriptors
.filter(Boolean)
.map(descriptor => new DocumentMergeConflict(document, descriptor));
}
if (value >= groupIndex) {
break;
}
private static completePartialMergeDescriptor(partial: IPartialMergeConflictDescriptor): interfaces.IDocumentMergeConflictDescriptor | null {
// Assume that descriptor.current.header, descriptor.incoming.header and descriptor.spliiter
// have valid ranges, fill in content and total ranges from these parts.
start += match[value] !== undefined ? match[value].length : 0;
if (!partial.currentHeader || !partial.splitter || !partial.incomingFooter) {
return null;
}
const groupMatch = match[groupIndex];
let targetMatchLength = groupMatch !== undefined ? groupMatch.length : -1;
let end = (start + targetMatchLength);
if (groupMatch !== undefined) {
// Move the end up if it's capped by a trailing \r\n, this is so regions don't expand into
// the line below, and can be "pulled down" by editing the line below
if (match[groupIndex].lastIndexOf('\n') === targetMatchLength - 1) {
end--;
return {
current: {
header: partial.currentHeader,
// Current content is range between header and splitter
content: new vscode.Range(partial.currentHeader.end, partial.splitter.start),
name: ''
},
splitter: partial.splitter,
incoming: {
header: partial.incomingFooter,
// Incoming content is range between splitter and footer
content: new vscode.Range(partial.splitter.end, partial.incomingFooter.end),
name: ''
},
// Entire range is between current header start and incoming header end
range: new vscode.Range(partial.currentHeader.start, partial.incomingFooter.end)
};
}
// .. for windows encodings of new lines
if (match[groupIndex].lastIndexOf('\r') === targetMatchLength - 2) {
end--;
}
}
static containsConflict(document: vscode.TextDocument): boolean {
if (!document) {
return false;
}
return new vscode.Range(document.positionAt(start), document.positionAt(end));
// TODO: Ask source control if the file contains a conflict
let text = document.getText();
return text.includes('<<<<<<<') && text.includes('>>>>>>>');
}
}
extensions/merge-conflict/src/mergeConflictParserLegacy.ts 0 → 100644
浏览文件 @ 76b6b177
/*---------------------------------------------------------------------------------------------
* Copyright (c) Microsoft Corporation. All rights reserved.
* Licensed under the MIT License. See License.txt in the project root for license information.
*--------------------------------------------------------------------------------------------*/
import * as vscode from 'vscode';
import * as interfaces from './interfaces';
import { DocumentMergeConflict } from './documentMergeConflict';
import * as vm from 'vm';
export class MergeConflictParser {
static scanDocument(document: vscode.TextDocument): interfaces.IDocumentMergeConflict[] {
// Conflict matching regex, comments are in the format of "description - [group index] group name"
// Premise is: Match the current change (<<<<<<), match anything up to the splitter (======) then
// match anything up to the incoming change (>>>>>>), this leaves some oddities with newlines not being
// pulled into the "body" of each change, DocumentMergeConflict.applyEdit will deal with these cases
// and append newlines when needed
const conflictMatcher = new RegExp([
/(^<<<<<<<\s(.+)\r?\n)/, // "Current" conflict header - [1] entire line, [2] name
/([\s\S]*?)/, // "Current" conflict body - [3] body text
/(^=======\r?\n)/, // Splitter - [4] entire line
/([\s\S]*?)/, // Incoming conflict body - [5]
/(^>>>>>>>\s(.+)\r?\n)/ // Incoming conflict header - [6] entire line, [7] name
].map(r => r.source).join(''), 'mg');
const offsetGroups = [1, 3, 4, 5, 6]; // Skip inner matches when calculating length
let text = document.getText();
let sandboxScope = {
result: [],
conflictMatcher,
text: text
};
const context = vm.createContext(sandboxScope);
const script = new vm.Script(`
let match;
while (match = conflictMatcher.exec(text)) {
// Ensure we don't get stuck in an infinite loop
if (match.index === conflictMatcher.lastIndex) {
conflictMatcher.lastIndex++;
}
result.push(match);
}`);
try {
// If the regex takes longer than 1s consider it dead
script.runInContext(context, { timeout: 1000 });
}
catch (ex) {
return [];
}
return sandboxScope.result.map(match => new DocumentMergeConflict(document, MergeConflictParser.matchesToDescriptor(document, match, offsetGroups)));
}
static containsConflict(document: vscode.TextDocument): boolean {
if (!document) {
return false;
}
// TODO: Ask source control if the file contains a conflict
let text = document.getText();
return text.includes('<<<<<<<') && text.includes('>>>>>>>');
}
static matchesToDescriptor(document: vscode.TextDocument, match: RegExpExecArray, offsets?: number[]): interfaces.IDocumentMergeConflictDescriptor {
var item: interfaces.IDocumentMergeConflictDescriptor = {
range: new vscode.Range(document.positionAt(match.index), document.positionAt(match.index + match[0].length)),
current: {
name: match[2],
header: this.getMatchPositions(document, match, 1, offsets),
content: this.getMatchPositions(document, match, 3, offsets),
},
splitter: this.getMatchPositions(document, match, 4, offsets),
incoming: {
name: match[9],
header: this.getMatchPositions(document, match, 6, offsets),
content: this.getMatchPositions(document, match, 5, offsets),
}
};
return item;
}
static getMatchPositions(document: vscode.TextDocument, match: RegExpExecArray, groupIndex: number, offsetGroups?: number[]): vscode.Range {
// Javascript doesnt give of offsets within the match, we need to calculate these
// based of the prior groups, skipping nested matches (yuck).
if (!offsetGroups) {
offsetGroups = match.map((i, idx) => idx);
}
let start = match.index;
for (var i = 0; i < offsetGroups.length; i++) {
let value = offsetGroups[i];
if (value >= groupIndex) {
break;
}
start += match[value] !== undefined ? match[value].length : 0;
}
const groupMatch = match[groupIndex];
let targetMatchLength = groupMatch !== undefined ? groupMatch.length : -1;
let end = (start + targetMatchLength);
if (groupMatch !== undefined) {
// Move the end up if it's capped by a trailing \r\n, this is so regions don't expand into
// the line below, and can be "pulled down" by editing the line below
if (match[groupIndex].lastIndexOf('\n') === targetMatchLength - 1) {
end--;
// .. for windows encodings of new lines
if (match[groupIndex].lastIndexOf('\r') === targetMatchLength - 2) {
end--;
}
}
}
return new vscode.Range(document.positionAt(start), document.positionAt(end));
}
}
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或